Method and Apparatus for Detecting Rogue Trading Activity

Abstract

A system and method for detecting fraudulent activity in the execution of transactions is disclosed. The system comprises a monitoring device for reviewing data relating to execution of transactions, a transaction profile and an alert module. The transaction profile includes a plurality of historic data items relating to typical transactions, which can be compared with current execution of transactions to generate an alert by the alert module if unusual activity is determined.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

The present application is a continuation of U.S. patent application Ser. No. 14/180,723, filed Feb. 14, 2014, which is hereby incorporated by reference in its entirety. The present application is also related to U.S. patent application Ser. No. 12/965,226, filed Dec. 10, 2010, and to U.K. Patent Application No. GB 1316319.1, filed Sep. 13, 2013, both of which are herein incorporated by reference in their entirety.

FIELD

This application relates to an apparatus, a method and a computer program product for detection of rogue trading.

BACKGROUND

A rogue trader is an individual who acts recklessly and independent of fellow employees to make unauthorized trades on behalf of their employer. The term is most often applicable to financial trading, in which professional traders make and approve financial transactions. Rogue traders typically trade in high-risk investments, which may cause considerable losses. These losses may be preceded by large, but unsustainable, profits. Financial institutions have, in the past, tended to largely ignore this potentially catastrophic form of operational risk, as described in the paper ‘A Review of Operational Risk in Banks and its Role in the Financial Crisis’, by De Jongh et al. and published in the SAJEMS NS 16 [2013] No. 4: 364-382.

An article on the Forbes website ‘Corporate Culture is Key element in Fighting Rogue Trading’, by Steve Culp, dated 21 Jun. 2012 notes that detection and prevention of rogue trading activity is not easy. Large financial institutions have made significant investments in sophisticated systems to identify unauthorized activities. The ability of individuals acting as rogue traders to avoid detection and to continue to make unauthorized trade remains a concern for risk management and executive teams in many leading organizations.

Current attempts at combatting rogue trading rely on rule-based systems in which transaction relating to trades are analyzed and, if the transactions trigger one or more of the rules, then the transaction is either not performed or an alert triggered. U.S. Pat. No. 6,965,886 [Goovrin et al., assigned to Actimize Limited Ltd, Petach Tikva, Israel] teaches such a system and method for collecting, filtering, analyzing, distributing and effectively utilizing highly relevant events in real time from huge quantities of data.

U.S. Pat. No. 8,145,560 [Kulkarni et al., assigned to FMR, Boston, Mass.] also teaches a rule-based system for detecting fraudulent activity on a network.

The known prior art methods for detection of fraudulent activity rely on a rule-based system. It is, however, established that rogue traders can circumvent the rules in the rule-based system if the traders understand how the rules operate. For example, a rogue trader may break down one trade into a number of smaller trades executed through different brokers in order to circumvent any rules restricting the size of a trade. Clearly it would be possible to construct a rule to avoid such circumvention once the issue has become known. There is, however, a need to develop a system, which can be used to identify potentially rogue trades at an earlier stage.

SUMMARY

A system and a method for detecting fraudulent activity are taught in this disclosure. The system comprises a monitoring device for reviewing any data relating to execution of transactions, such as trades, as well as an alert module for generating an alert on detection of irregularities in the transaction. The monitoring device is able to access a transaction profile, which includes a plurality of historic data items relating to typical transactions. The transaction profile is used to establish a profile base line for the execution of transactions, which are allowed by a financial institution's internal rules. The data items in the transaction profile could include but are not limited to sales transactions, put options, sale options, address lookup through a DNS server, network addresses, routing information, time, date and device cookies stored on a trader's computer. These data items taken alone may not indicate that an unauthorized transaction has taken place. However, the data items enable the transaction profile to be constructed for a trader carrying out authorized trades. Should any one of the traders in the financial institution deviate from the constructed transaction profile, then the pattern of trades can be investigated manually by, for example, a compliance officer in order to determine whether the trader is acting in accordance with the bank's accepted guidelines. The alert module generates an alert generally in real-time or at a later date on detection of irregularities or deviations form the transaction profile. The irregularities or deviation may not indicate directly that one of the traders is a rogue trader, but will allow the investigation by a financial institution's compliance department to ensure that no issue is raised.

In one aspect of the invention, the system may include a number of transaction profiles indicating different skills or responsibility levels of the traders. For example, a junior trader may only be able to undertake certain transactions and/or execute a certain level of transactions. A more experienced trader will be empowered to execute different types of transactions. It would be expected that the transaction profiles for the different skill levels of the traders will therefore be different. A group transaction profile can be established for a particular group of traders.

The method for detecting fraudulent activity involves the receipt from a first device such as a trader's computer, phone or tablet, and comparing the data items with a transaction profile. The alert will be generated and can then be passed to a supervisor for action.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 shows an overview of a user computer connected to a network.

FIG. 2 shows a flow diagram of the method.

DETAILED DESCRIPTION

The invention will now be described on the basis of the drawings. It will be understood that the embodiments and aspects of the invention described herein are only examples and do not limit the protective scope of the claims in any way. The invention is defined by the claims and their equivalents. It will be understood that features of one aspect or embodiment of the invention can be combined with a feature of a different aspect or aspects and/or embodiments of the invention.

FIG. 1 shows a system 100 for detecting fraudulent activity according to one aspect of the invention. The system includes a user computer 10 with a plurality of outgoing connections 20 and a plurality of incoming connections 30. The user computer 10 could alternatively be a smartphone or tablet as well as a desktop or laptop computer and the term is intended to cover these alternates. The outgoing connections 20 and the incoming connections 30 are connected to one or more transactional servers 15a or 15b. The transmission of data over the outgoing connections 20 and the incoming connections 30 in the system 100 is carried out using, for example, packet-based networks or circuit-based networks. Packet-based networks include, but are not limited to, the Internet, a private internet protocol (IP) network, or a wireless network (such as GPRS, UMTS or LTE). Circuit-based networks include, but are not limited to, the public switched telephone network (PSTN), a wireless network (such as RAN Bluetooth, CDMA, TDMA or GSM).

A user at the user computer 10 is a stock trader, who is engaged in the buying and selling of financial products, such as bonds, stocks, futures, derivatives and shares, in for example the stock exchange (bourse) or commodity markets. The trader interacts with the user computer 10 to retrieve information about financial products from information sources, such as Reuters or Bloomberg through a webserver 16 linked to the Internet 17 or an intranet, and to execute trades in the financial products. The trades include buy and sell orders. The buy and sell orders can be carried out using a program running on a transactional server 15, using a browser program on the user computer 10 or by using a dedicated trading application program running on the user computer 10. The user computer 10 can include a verification device 12 for verifying the identity of the trader wishing to execute the trade, and the authenticity of the trade. Suitable verification devices 12 include, but are not limited to, card readers, one-time key devices, password entry or a combination thereof. The verification device 12 includes an alert module 67 for generating an alert in the event that the trade is not authorized. Messages 70 are passed between the user computer 10 and the transaction servers 15a-b to instruct the transaction servers 15a-b to execute the trades.

The transaction server 15a can also include an authorization device 19, which reviews the transaction before execution for compliance with the financial institution's compliance rules. The authorization device 19 is, for example, a rules-based device, as known in the art. The authorization device 19 is shown only in the transaction server 15a. The authorization device 19 could be on all of the transaction servers 15a-b or on a single transaction server 15a, but accessed by other ones of the transaction servers 15a.

A monitoring device 60 installed within the system 100 in which the user computer 10 and the transactional servers 15 are operating. The monitoring device 60 continually monitors the system 100 and the user computer 10 as well as the messages 70 exchanged within the system 100 and/or generated by the user computer 10 and/or generated by the transaction server B. The monitoring device 60 uses a variety of data and data sources for performing the monitoring.

Examples of the data collected by the monitoring device 60 include data items 66 relating to the websites accessed by the trader. These data items include but are not limited to global address lookup (i.e., global IP address used by ICAAN), network address, routing information, time, date, device cookies downloaded or used, device fingerprints as well as volume and timing of trades and combinations thereof. The monitoring device 60 can also use data sources based on network flow traffic statistics through the system 100 and to the outside network. These data sources include proxy logs and NetFlow records, which record the destination of data sent through outgoing ports 25 of the user computer 10 and/or the transactional server 15 as well as the source of incoming data received through the incoming ports 35 at the user computer 10 and/or the transactional server 15.

Many computer networks 130 also have a DNS server 110 located in a private network, as well as having access to public DNS servers 120. The DNS server 110 includes a variety of data log entries, including time stamps, indicating which ones of the trader computers attempted to access which web sites or external servers at which period of time.

The monitoring device 60 can also review headers in emails and/or other messages 70 sent throughout the computer network 130. The email headers will include information, such as the time, the destination and the source, as well as having information about the size of the email.

It will be appreciated that these data sources are merely exemplary and that other data sources can be used or supplied. Only a single user computer 10 is shown in FIG. 1 for simplicity. In practice, there will be a large number of user computers 10 and transactional servers 15. It will be appreciated that the system 100 may also contain other devices that can generate messages 70 or other data.

The monitoring device 60 creates a transaction profile 62 for the traders/users and/or the user computer 10 based on behaviour of the user or a group of users. The transaction profile 62 is stored in a user profile database 65 attached to the monitoring device 60. It will be appreciated that the user profile database 65 contains more than one transaction profile 62 and that each one of the traders (and possibly the user computers 10) can have a separate transaction profile 62 and/or a group transaction profile 62 can be created for a group of traders. The transaction profile 62 in the user profile database 65 receives the data items 66 that indicate how the user computer 10 generally reacts with the system 100 as well as with the transactional servers 15 and other devices in the system 100 such as the web server 16 or the DNS server 110. The user profile database 62 will continually be updated as new ones of the data items 66 relating to activity of the user computer 10 are generated. The transaction profile 62 creates in essence a baseline from which a ‘normal’ behavior of the trader and/or the user computer 10 can be deduced.

The transaction profile 62 will be built up from the trader's typical interactions. For example, a trader will typically make trades at a particular rate, volume and price. The traders will also potentially typically trade in particular markets or stocks and exhibit periodic and seasonal variances. The transaction profile 62 is therefore a statistical model of the trader based on these trades. The transaction profile 62 represents the probability of a specific interaction based on historical behavior and is thus a baseline against which rogue trading activities can be detected.

The transaction profile 62 for a group of traders can be built up to represent the baseline for a particular trading activity. One example would be a group of commodity traders. The user profile database 65 could have both the individual transaction profile 62 for an individual one of the traders and a further group transaction profile for the group of commodity traders.

Suppose now that a trader (or a person using the user computer 10) initiates the rogue trading activities. The monitoring device 60 will receive further data items 66 relating to the rogue behavior. The further data items 66 indicate that behavior of the user computer 10 deviates from the normal baseline behaviour expected by comparison with the transaction profile 62 stored in the user profile database 65. Non limiting examples of such deviant behaviour include massive amounts of data being transferred to one of the transaction servers 15, unusual patterns of orders or access to a rarely used website or automated execution of trades. The monitoring device 60 can notify in real time an administrator of a possible fraudulent activity initiated by the user computer 10 and the administrator or compliance department can investigate the activities from the user computer 10.

FIG. 2 shows an outline of a process 200 for the detection of fraudulent activity by the rogue trader. The process 200 starts at 210 by a log in of the trader at the user computer 10 and in step 220 the data items 66 from the various data sources are gathered. In step 230 the gathered data items 66 are compared with one or more of the transaction profiles 62. There may be more than one transaction profile 62 and, in this case, the transaction profile 62 most representative of the trader at the user computer 10 is chosen. If an anomaly or irregularity is discovered, an alert is generated in step 240 such that the administrator or the compliance officer can investigate in step 250. In step 260, the transaction profile 62 is updated from the newly gathered data items 66.

The updating of the transaction profile 62 in step 260 ensures that the transaction profile 62 is continually adapted to new types of orders or data sources or other transactional servers 15 inserted into the system 100.

In a further aspect of the invention, the system and method can be used to detect further types of fraudulent behaviour using the user computer 10 or in the system 100. It would be possible, for example, to use the teachings of a disclosure to detect other types of fraud by users of the user computer 10. The fraud can be detected by, for example, identifying anomalous attempts to access certain websites, which are not normally accessed, or, by an attempt to transfer significant amounts of data to a computer or memory device that is not normally in use, or by the generation of a large number of emails in a particular period of time.

The detection of fraud is made by detection of unusual activity in the transaction profile 62. One further method for identifying fraud is by comparing the different ones of the transaction profiles 62 of different users of the computer 10. If one of the transaction profiles 62 is substantially different than other ones of the transaction profiles 62, then notification can be made to an administrator or a compliance officer to investigate activities of the trader and the user computer 10. Another method for identifying fraud is if the transaction profile 62 suddenly changes or is unexpectedly modified. Similarly if the activity at the user computer 10 does not reflect the transaction profile for the expected group transaction profile 62, then the notification can also be issued.

The transaction profile 62 can be initially created by using a set of training data obtained from analysing over a period of time. The training data will represent real-life, non-rogue situations.

REFERENCE NUMERALS

• 10 User Computer
• 12 Verification Device
• 15 Transactional Servers
• 16 Web server
• 17 Internet
• 19 Authorization Device
• 20 Outgoing Connections
• 25 Outgoing port
• 30 Incoming Connections
• 35 Incoming port
• 60 Monitoring Device
• 62 Transaction Profile
• 65 User profile database
• 66 Data Items
• 67 Alert module
• 70 Message
• 100 System
• 110 Private DNS Server
• 120 Public DNS Server
• 130 Computer Network

Claims

1. A system for detecting fraudulent activity of at least one user, the system comprising:

a plurality of transactional devices connected to a monitoring device, wherein the plurality of transactional devices are configured to execute transactions, and wherein the monitoring device is configured to create at least two transaction baseline profiles of at least two different users executing transactions on at least two of the plurality of transactional devices, wherein the at least two transaction baseline profiles include a plurality of data items accruing from the historical behavior of each of the at least two users, wherein the monitoring device is configured to continuously receive further data items from the plurality of transactional devices relating to the present behavior of at least two users for adapting the at least two transaction baseline profiles; and

an alert module for generating an alert on detection of irregularities between the plurality of data items accruing from the at least two transaction baseline profiles and a plurality of data items relating the present behavior of each of the at least two users by the monitoring device, wherein the monitoring device is configured to compare transaction baseline profiles of the same user and of at least two different users.

2. The system of claim 1, wherein the data items comprise at least one of: security purchase transaction data, security sale transaction data, put options pricing data, global address lookup, network address, routing information, transaction time stamp, transaction date stamp, and device cookies data.

3. The system of claim 1, further including a plurality of transaction profiles for a plurality of traders having different trading authorization levels or trader groups.

4. The system of claim 1, wherein the historic data items comprise data items relating to at least one of a rate, volume, price, stock ticker, periodic variances in volume rates, and seasonal variances in rates or volumes.

5. A method for detecting fraudulent activity of at least one user, the method comprising:

receiving from a plurality of transactional devices a plurality of data items;

creating at least two transaction baseline profiles of at least two different users executing transactions on at least two of the plurality of transactional devices, wherein the at least two transaction baseline profiles including the plurality of data items accruing from the historical behavior of the at least two different users by the monitoring device;

adapting the at least two transaction baseline profiles of at least two different users including the plurality of data items accruing from the present behavior of at least two different users by the monitoring device;

comparing the received plurality of data items accruing from the present behavior of at least two different users with the transaction baseline profile of the same user and of the at least one other user of the at least two different users by the monitoring device; and

generating an alert from the monitoring device on detection of irregularities between the plurality of data items accruing from the at least two transaction baseline profiles and a plurality of data items relating the present behavior of each of the at least two different users.

6. The method of claim 5, further comprising updating the transaction baseline profile with the received data items over a period of time.

7. A computer program product, tangibly embodied in a non-transitory computer readable storage medium, for detecting fraudulent activity, the computer program product including instructions operable to cause a data processing apparatus to:

receive from a plurality of transactional devices a plurality of data items;

create at least two transaction baseline profiles of at least two different users executing transactions on at least two of the plurality of transactional devices, wherein the at least two transaction baseline profiles include a plurality of data items accruing from historical behavior of the at least two different users by a monitoring device;

adapt that at least two transaction baseline profiles of at least two different users including the plurality of data items accruing from the present behavior of at least two different users by the monitoring device;

compare the received plurality of data items accruing from the present behavior of at least one two different users by the monitoring device; and

generate an alert from the monitoring device on detection of irregularities between the plurality of data items accruing from the at least two transaction baseline profiles and a plurality of data items relating the present behavior of each of the at least two users.